]> code.delx.au - gnu-emacs/blob - lisp/net/shr.el
Render empty <ul><li><ul> correctly
[gnu-emacs] / lisp / net / shr.el
1 ;;; shr.el --- Simple HTML Renderer
2
3 ;; Copyright (C) 2010-2016 Free Software Foundation, Inc.
4
5 ;; Author: Lars Magne Ingebrigtsen <larsi@gnus.org>
6 ;; Keywords: html
7
8 ;; This file is part of GNU Emacs.
9
10 ;; GNU Emacs is free software: you can redistribute it and/or modify
11 ;; it under the terms of the GNU General Public License as published by
12 ;; the Free Software Foundation, either version 3 of the License, or
13 ;; (at your option) any later version.
14
15 ;; GNU Emacs is distributed in the hope that it will be useful,
16 ;; but WITHOUT ANY WARRANTY; without even the implied warranty of
17 ;; MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
18 ;; GNU General Public License for more details.
19
20 ;; You should have received a copy of the GNU General Public License
21 ;; along with GNU Emacs. If not, see <http://www.gnu.org/licenses/>.
22
23 ;;; Commentary:
24
25 ;; This package takes a HTML parse tree (as provided by
26 ;; libxml-parse-html-region) and renders it in the current buffer. It
27 ;; does not do CSS, JavaScript or anything advanced: It's geared
28 ;; towards rendering typical short snippets of HTML, like what you'd
29 ;; find in HTML email and the like.
30
31 ;;; Code:
32
33 (eval-when-compile (require 'cl))
34 (eval-when-compile (require 'url)) ;For url-filename's setf handler.
35 (require 'browse-url)
36 (require 'subr-x)
37 (require 'dom)
38 (require 'seq)
39 (require 'svg)
40
41 (defgroup shr nil
42 "Simple HTML Renderer"
43 :version "25.1"
44 :group 'web)
45
46 (defcustom shr-max-image-proportion 0.9
47 "How big pictures displayed are in relation to the window they're in.
48 A value of 0.7 means that they are allowed to take up 70% of the
49 width and height of the window. If they are larger than this,
50 and Emacs supports it, then the images will be rescaled down to
51 fit these criteria."
52 :version "24.1"
53 :group 'shr
54 :type 'float)
55
56 (defcustom shr-blocked-images nil
57 "Images that have URLs matching this regexp will be blocked."
58 :version "24.1"
59 :group 'shr
60 :type '(choice (const nil) regexp))
61
62 (defcustom shr-use-fonts t
63 "If non-nil, use proportional fonts for text."
64 :version "25.1"
65 :group 'shr
66 :type 'boolean)
67
68 (defcustom shr-use-colors t
69 "If non-nil, respect color specifications in the HTML."
70 :version "25.2"
71 :group 'shr
72 :type 'boolean)
73
74 (defcustom shr-table-horizontal-line nil
75 "Character used to draw horizontal table lines.
76 If nil, don't draw horizontal table lines."
77 :group 'shr
78 :type '(choice (const nil) character))
79
80 (defcustom shr-table-vertical-line ?\s
81 "Character used to draw vertical table lines."
82 :group 'shr
83 :type 'character)
84
85 (defcustom shr-table-corner ?\s
86 "Character used to draw table corners."
87 :group 'shr
88 :type 'character)
89
90 (defcustom shr-hr-line ?-
91 "Character used to draw hr lines."
92 :group 'shr
93 :type 'character)
94
95 (defcustom shr-width nil
96 "Frame width to use for rendering.
97 May either be an integer specifying a fixed width in characters,
98 or nil, meaning that the full width of the window should be
99 used."
100 :version "25.1"
101 :type '(choice (integer :tag "Fixed width in characters")
102 (const :tag "Use the width of the window" nil))
103 :group 'shr)
104
105 (defcustom shr-bullet "* "
106 "Bullet used for unordered lists.
107 Alternative suggestions are:
108 - \" \"
109 - \" \""
110 :version "24.4"
111 :type 'string
112 :group 'shr)
113
114 (defcustom shr-external-browser 'browse-url-default-browser
115 "Function used to launch an external browser."
116 :version "24.4"
117 :group 'shr
118 :type 'function)
119
120 (defcustom shr-image-animate t
121 "Non nil means that images that can be animated will be."
122 :version "24.4"
123 :group 'shr
124 :type 'boolean)
125
126 (defvar shr-content-function nil
127 "If bound, this should be a function that will return the content.
128 This is used for cid: URLs, and the function is called with the
129 cid: URL as the argument.")
130
131 (defvar shr-put-image-function 'shr-put-image
132 "Function called to put image and alt string.")
133
134 (defface shr-strike-through '((t (:strike-through t)))
135 "Font for <s> elements."
136 :group 'shr)
137
138 (defface shr-link
139 '((t (:inherit link)))
140 "Font for link elements."
141 :group 'shr)
142
143 (defvar shr-inhibit-images nil
144 "If non-nil, inhibit loading images.")
145
146 (defvar shr-external-rendering-functions nil
147 "Alist of tag/function pairs used to alter how shr renders certain tags.
148 For instance, eww uses this to alter rendering of title, forms
149 and other things:
150 ((title . eww-tag-title)
151 (form . eww-tag-form)
152 ...)")
153
154 ;;; Internal variables.
155
156 (defvar shr-folding-mode nil)
157 (defvar shr-start nil)
158 (defvar shr-indentation 0)
159 (defvar shr-internal-width nil)
160 (defvar shr-list-mode nil)
161 (defvar shr-content-cache nil)
162 (defvar shr-kinsoku-shorten nil)
163 (defvar shr-table-depth 0)
164 (defvar shr-stylesheet nil)
165 (defvar shr-base nil)
166 (defvar shr-depth 0)
167 (defvar shr-warning nil)
168 (defvar shr-ignore-cache nil)
169 (defvar shr-target-id nil)
170 (defvar shr-table-separator-length 1)
171 (defvar shr-table-separator-pixel-width 0)
172 (defvar shr-table-id nil)
173 (defvar shr-current-font nil)
174 (defvar shr-internal-bullet nil)
175
176 (defvar shr-map
177 (let ((map (make-sparse-keymap)))
178 (define-key map "a" 'shr-show-alt-text)
179 (define-key map "i" 'shr-browse-image)
180 (define-key map "z" 'shr-zoom-image)
181 (define-key map [?\t] 'shr-next-link)
182 (define-key map [?\M-\t] 'shr-previous-link)
183 (define-key map [follow-link] 'mouse-face)
184 (define-key map [mouse-2] 'shr-browse-url)
185 (define-key map "I" 'shr-insert-image)
186 (define-key map "w" 'shr-copy-url)
187 (define-key map "u" 'shr-copy-url)
188 (define-key map "v" 'shr-browse-url)
189 (define-key map "O" 'shr-save-contents)
190 (define-key map "\r" 'shr-browse-url)
191 map))
192
193 (defvar shr-image-map
194 (let ((map (copy-keymap shr-map)))
195 (when (boundp 'image-map)
196 (set-keymap-parent map image-map))
197 map))
198
199 ;; Public functions and commands.
200 (declare-function libxml-parse-html-region "xml.c"
201 (start end &optional base-url discard-comments))
202
203 (defun shr-render-buffer (buffer)
204 "Display the HTML rendering of the current buffer."
205 (interactive (list (current-buffer)))
206 (or (fboundp 'libxml-parse-html-region)
207 (error "This function requires Emacs to be compiled with libxml2"))
208 (pop-to-buffer "*html*")
209 (erase-buffer)
210 (shr-insert-document
211 (with-current-buffer buffer
212 (libxml-parse-html-region (point-min) (point-max))))
213 (goto-char (point-min)))
214
215 ;;;###autoload
216 (defun shr-render-region (begin end &optional buffer)
217 "Display the HTML rendering of the region between BEGIN and END."
218 (interactive "r")
219 (unless (fboundp 'libxml-parse-html-region)
220 (error "This function requires Emacs to be compiled with libxml2"))
221 (with-current-buffer (or buffer (current-buffer))
222 (let ((dom (libxml-parse-html-region begin end)))
223 (delete-region begin end)
224 (goto-char begin)
225 (shr-insert-document dom))))
226
227 (defun shr--have-one-fringe-p ()
228 "Return non-nil if we know at least one of the fringes has non-zero width."
229 (and (fboundp 'fringe-columns)
230 (or (not (zerop (fringe-columns 'right)))
231 (not (zerop (fringe-columns 'left))))))
232
233 ;;;###autoload
234 (defun shr-insert-document (dom)
235 "Render the parsed document DOM into the current buffer.
236 DOM should be a parse tree as generated by
237 `libxml-parse-html-region' or similar."
238 (setq shr-content-cache nil)
239 (let ((start (point))
240 (shr-start nil)
241 (shr-base nil)
242 (shr-depth 0)
243 (shr-table-id 0)
244 (shr-warning nil)
245 (shr-table-separator-pixel-width (shr-string-pixel-width "-"))
246 (shr-internal-bullet (cons shr-bullet
247 (shr-string-pixel-width shr-bullet)))
248 (shr-internal-width (or (and shr-width
249 (if (not shr-use-fonts)
250 shr-width
251 (* shr-width (frame-char-width))))
252 ;; We need to adjust the available
253 ;; width for when the user disables
254 ;; the fringes, which will cause the
255 ;; display engine usurp one column for
256 ;; the continuation glyph.
257 (if (not shr-use-fonts)
258 (- (window-body-width) 1
259 (if (and (null shr-width)
260 (not (shr--have-one-fringe-p)))
261 0
262 1))
263 (- (window-body-width nil t)
264 (* 2 (frame-char-width))
265 (if (and (null shr-width)
266 (not (shr--have-one-fringe-p)))
267 (* (frame-char-width) 2)
268 0)))))
269 bidi-display-reordering)
270 ;; If the window was hscrolled for some reason, shr-fill-lines
271 ;; below will misbehave, because it silently assumes that it
272 ;; starts with a non-hscrolled window (vertical-motion will move
273 ;; to a wrong place otherwise).
274 (set-window-hscroll nil 0)
275 (shr-descend dom)
276 (shr-fill-lines start (point))
277 (shr--remove-blank-lines-at-the-end start (point))
278 (when shr-warning
279 (message "%s" shr-warning))))
280
281 (defun shr--remove-blank-lines-at-the-end (start end)
282 (save-restriction
283 (save-excursion
284 (current-buffer)
285 (narrow-to-region start end)
286 (goto-char end)
287 (when (and (re-search-backward "[^ \n]" nil t)
288 (not (eobp)))
289 (forward-char 1)
290 (delete-region (point) (point-max))
291 (unless (bolp)
292 (insert "\n"))))))
293
294 (defun shr-copy-url (&optional image-url)
295 "Copy the URL under point to the kill ring.
296 If IMAGE-URL (the prefix) is non-nil, or there is no link under
297 point, but there is an image under point then copy the URL of the
298 image under point instead.
299 If called twice, then try to fetch the URL and see whether it
300 redirects somewhere else."
301 (interactive "P")
302 (let ((url (or (get-text-property (point) 'shr-url)
303 (get-text-property (point) 'image-url))))
304 (cond
305 ((not url)
306 (message "No URL under point"))
307 ;; Resolve redirected URLs.
308 ((equal url (car kill-ring))
309 (url-retrieve
310 url
311 (lambda (a)
312 (when (and (consp a)
313 (eq (car a) :redirect))
314 (with-temp-buffer
315 (insert (cadr a))
316 (goto-char (point-min))
317 ;; Remove common tracking junk from the URL.
318 (when (re-search-forward ".utm_.*" nil t)
319 (replace-match "" t t))
320 (message "Copied %s" (buffer-string))
321 (copy-region-as-kill (point-min) (point-max)))))
322 nil t))
323 ;; Copy the URL to the kill ring.
324 (t
325 (with-temp-buffer
326 (insert (url-encode-url url))
327 (copy-region-as-kill (point-min) (point-max))
328 (message "Copied %s" (buffer-string)))))))
329
330 (defun shr-next-link ()
331 "Skip to the next link."
332 (interactive)
333 (let ((current (get-text-property (point) 'shr-url))
334 (start (point))
335 skip)
336 (while (and (not (eobp))
337 (equal (get-text-property (point) 'shr-url) current))
338 (forward-char 1))
339 (cond
340 ((and (not (eobp))
341 (get-text-property (point) 'shr-url))
342 ;; The next link is adjacent.
343 (message "%s" (get-text-property (point) 'help-echo)))
344 ((or (eobp)
345 (not (setq skip (text-property-not-all (point) (point-max)
346 'shr-url nil))))
347 (goto-char start)
348 (message "No next link"))
349 (t
350 (goto-char skip)
351 (message "%s" (get-text-property (point) 'help-echo))))))
352
353 (defun shr-previous-link ()
354 "Skip to the previous link."
355 (interactive)
356 (let ((start (point))
357 (found nil))
358 ;; Skip past the current link.
359 (while (and (not (bobp))
360 (get-text-property (point) 'help-echo))
361 (forward-char -1))
362 ;; Find the previous link.
363 (while (and (not (bobp))
364 (not (setq found (get-text-property (point) 'help-echo))))
365 (forward-char -1))
366 (if (not found)
367 (progn
368 (message "No previous link")
369 (goto-char start))
370 ;; Put point at the start of the link.
371 (while (and (not (bobp))
372 (get-text-property (point) 'help-echo))
373 (forward-char -1))
374 (forward-char 1)
375 (message "%s" (get-text-property (point) 'help-echo)))))
376
377 (defun shr-show-alt-text ()
378 "Show the ALT text of the image under point."
379 (interactive)
380 (let ((text (get-text-property (point) 'shr-alt)))
381 (if (not text)
382 (message "No image under point")
383 (message "%s" (shr-fill-text text)))))
384
385 (defun shr-browse-image (&optional copy-url)
386 "Browse the image under point.
387 If COPY-URL (the prefix if called interactively) is non-nil, copy
388 the URL of the image to the kill buffer instead."
389 (interactive "P")
390 (let ((url (get-text-property (point) 'image-url)))
391 (cond
392 ((not url)
393 (message "No image under point"))
394 (copy-url
395 (with-temp-buffer
396 (insert url)
397 (copy-region-as-kill (point-min) (point-max))
398 (message "Copied %s" url)))
399 (t
400 (message "Browsing %s..." url)
401 (browse-url url)))))
402
403 (defun shr-insert-image ()
404 "Insert the image under point into the buffer."
405 (interactive)
406 (let ((url (get-text-property (point) 'image-url)))
407 (if (not url)
408 (message "No image under point")
409 (message "Inserting %s..." url)
410 (url-retrieve url 'shr-image-fetched
411 (list (current-buffer) (1- (point)) (point-marker))
412 t t))))
413
414 (defun shr-zoom-image ()
415 "Toggle the image size.
416 The size will be rotated between the default size, the original
417 size, and full-buffer size."
418 (interactive)
419 (let ((url (get-text-property (point) 'image-url))
420 (size (get-text-property (point) 'image-size))
421 (buffer-read-only nil))
422 (if (not url)
423 (message "No image under point")
424 ;; Delete the old picture.
425 (while (get-text-property (point) 'image-url)
426 (forward-char -1))
427 (forward-char 1)
428 (let ((start (point)))
429 (while (get-text-property (point) 'image-url)
430 (forward-char 1))
431 (forward-char -1)
432 (put-text-property start (point) 'display nil)
433 (when (> (- (point) start) 2)
434 (delete-region start (1- (point)))))
435 (message "Inserting %s..." url)
436 (url-retrieve url 'shr-image-fetched
437 (list (current-buffer) (1- (point)) (point-marker)
438 (list (cons 'size
439 (cond ((or (eq size 'default)
440 (null size))
441 'original)
442 ((eq size 'original)
443 'full)
444 ((eq size 'full)
445 'default)))))
446 t))))
447
448 ;;; Utility functions.
449
450 (defsubst shr-generic (dom)
451 (dolist (sub (dom-children dom))
452 (if (stringp sub)
453 (shr-insert sub)
454 (shr-descend sub))))
455
456 (defun shr-descend (dom)
457 (let ((function
458 (intern (concat "shr-tag-" (symbol-name (dom-tag dom))) obarray))
459 ;; Allow other packages to override (or provide) rendering
460 ;; of elements.
461 (external (cdr (assq (dom-tag dom) shr-external-rendering-functions)))
462 (style (dom-attr dom 'style))
463 (shr-stylesheet shr-stylesheet)
464 (shr-depth (1+ shr-depth))
465 (start (point)))
466 ;; shr uses many frames per nested node.
467 (if (> shr-depth (/ max-specpdl-size 15))
468 (setq shr-warning "Too deeply nested to render properly; consider increasing `max-specpdl-size'")
469 (when style
470 (if (string-match "color\\|display\\|border-collapse" style)
471 (setq shr-stylesheet (nconc (shr-parse-style style)
472 shr-stylesheet))
473 (setq style nil)))
474 ;; If we have a display:none, then just ignore this part of the DOM.
475 (unless (equal (cdr (assq 'display shr-stylesheet)) "none")
476 (cond (external
477 (funcall external dom))
478 ((fboundp function)
479 (funcall function dom))
480 (t
481 (shr-generic dom)))
482 (when (and shr-target-id
483 (equal (dom-attr dom 'id) shr-target-id))
484 ;; If the element was empty, we don't have anything to put the
485 ;; anchor on. So just insert a dummy character.
486 (when (= start (point))
487 (insert "*"))
488 (put-text-property start (1+ start) 'shr-target-id shr-target-id))
489 ;; If style is set, then this node has set the color.
490 (when style
491 (shr-colorize-region
492 start (point)
493 (cdr (assq 'color shr-stylesheet))
494 (cdr (assq 'background-color shr-stylesheet))))))))
495
496 (defun shr-fill-text (text)
497 (if (zerop (length text))
498 text
499 (with-temp-buffer
500 (let ((shr-indentation 0)
501 (shr-start nil)
502 (shr-internal-width (- (window-body-width nil t)
503 (* 2 (frame-char-width))
504 ;; Adjust the window width for when
505 ;; the user disables the fringes,
506 ;; which causes the display engine
507 ;; to usurp one column for the
508 ;; continuation glyph.
509 (if (and (null shr-width)
510 (not (shr--have-one-fringe-p)))
511 (* (frame-char-width) 2)
512 0))))
513 (shr-insert text)
514 (buffer-string)))))
515
516 (define-inline shr-char-breakable-p (char)
517 "Return non-nil if a line can be broken before and after CHAR."
518 (inline-quote (aref fill-find-break-point-function-table ,char)))
519 (define-inline shr-char-nospace-p (char)
520 "Return non-nil if no space is required before and after CHAR."
521 (inline-quote (aref fill-nospace-between-words-table ,char)))
522
523 ;; KINSOKU is a Japanese word meaning a rule that should not be violated.
524 ;; In Emacs, it is a term used for characters, e.g. punctuation marks,
525 ;; parentheses, and so on, that should not be placed in the beginning
526 ;; of a line or the end of a line.
527 (define-inline shr-char-kinsoku-bol-p (char)
528 "Return non-nil if a line ought not to begin with CHAR."
529 (inline-letevals (char)
530 (inline-quote (and (not (eq ,char ?'))
531 (aref (char-category-set ,char) ?>)))))
532 (define-inline shr-char-kinsoku-eol-p (char)
533 "Return non-nil if a line ought not to end with CHAR."
534 (inline-quote (aref (char-category-set ,char) ?<)))
535 (unless (shr-char-kinsoku-bol-p (make-char 'japanese-jisx0208 33 35))
536 (load "kinsoku" nil t))
537
538 (defun shr-pixel-column ()
539 (if (not shr-use-fonts)
540 (current-column)
541 (if (not (get-buffer-window (current-buffer)))
542 (save-window-excursion
543 (set-window-buffer nil (current-buffer))
544 (car (window-text-pixel-size nil (line-beginning-position) (point))))
545 (car (window-text-pixel-size nil (line-beginning-position) (point))))))
546
547 (defun shr-pixel-region ()
548 (- (shr-pixel-column)
549 (save-excursion
550 (goto-char (mark))
551 (shr-pixel-column))))
552
553 (defun shr-string-pixel-width (string)
554 (if (not shr-use-fonts)
555 (length string)
556 (with-temp-buffer
557 (insert string)
558 (shr-pixel-column))))
559
560 (defun shr-insert (text)
561 (when (and (not (bolp))
562 (get-text-property (1- (point)) 'image-url))
563 (insert "\n"))
564 (cond
565 ((eq shr-folding-mode 'none)
566 (let ((start (point)))
567 (insert text)
568 (save-restriction
569 (narrow-to-region start (point))
570 ;; Remove soft hyphens.
571 (goto-char (point-min))
572 (while (search-forward "­" nil t)
573 (replace-match "" t t))
574 (goto-char (point-max)))))
575 (t
576 (let ((font-start (point)))
577 (when (and (string-match "\\`[ \t\n\r ]" text)
578 (not (bolp))
579 (not (eq (char-after (1- (point))) ? )))
580 (insert " "))
581 (let ((start (point))
582 (bolp (bolp)))
583 (insert text)
584 (save-restriction
585 (narrow-to-region start (point))
586 (goto-char start)
587 (when (looking-at "[ \t\n\r ]+")
588 (replace-match "" t t))
589 (while (re-search-forward "[ \t\n\r ]+" nil t)
590 (replace-match " " t t))
591 ;; Remove soft hyphens.
592 (goto-char (point-min))
593 (while (search-forward "­" nil t)
594 (replace-match "" t t))
595 (goto-char (point-max)))
596 ;; We may have removed everything we inserted if if was just
597 ;; spaces.
598 (unless (= font-start (point))
599 ;; Mark all lines that should possibly be folded afterwards.
600 (when bolp
601 (shr-mark-fill start))
602 (when shr-use-fonts
603 (put-text-property font-start (point)
604 'face
605 (or shr-current-font 'variable-pitch)))))))))
606
607 (defun shr-fill-lines (start end)
608 (if (<= shr-internal-width 0)
609 nil
610 (save-restriction
611 (narrow-to-region start end)
612 (goto-char start)
613 (when (get-text-property (point) 'shr-indentation)
614 (shr-fill-line))
615 (while (setq start (next-single-property-change start 'shr-indentation))
616 (goto-char start)
617 (when (bolp)
618 (shr-fill-line)))
619 (goto-char (point-max)))))
620
621 (defun shr-vertical-motion (column)
622 (if (not shr-use-fonts)
623 (move-to-column column)
624 (unless (eolp)
625 (forward-char 1))
626 (vertical-motion (cons (/ column (frame-char-width)) 0))
627 (unless (eolp)
628 (forward-char 1))))
629
630 (defun shr-fill-line ()
631 (let ((shr-indentation (get-text-property (point) 'shr-indentation))
632 (continuation (get-text-property
633 (point) 'shr-continuation-indentation))
634 start)
635 (put-text-property (point) (1+ (point)) 'shr-indentation nil)
636 (let ((face (get-text-property (point) 'face))
637 (background-start (point)))
638 (shr-indent)
639 (when face
640 (put-text-property background-start (point) 'face
641 `,(shr-face-background face))))
642 (setq start (point))
643 (setq shr-indentation (or continuation shr-indentation))
644 (shr-vertical-motion shr-internal-width)
645 (when (looking-at " $")
646 (delete-region (point) (line-end-position)))
647 (while (not (eolp))
648 ;; We have to do some folding. First find the first
649 ;; previous point suitable for folding.
650 (if (or (not (shr-find-fill-point (line-beginning-position)))
651 (= (point) start))
652 ;; We had unbreakable text (for this width), so just go to
653 ;; the first space and carry on.
654 (progn
655 (beginning-of-line)
656 (skip-chars-forward " ")
657 (search-forward " " (line-end-position) 'move)))
658 ;; Success; continue.
659 (when (= (preceding-char) ?\s)
660 (delete-char -1))
661 (let ((face (get-text-property (point) 'face))
662 (background-start (point)))
663 (insert "\n")
664 (shr-indent)
665 (when face
666 (put-text-property background-start (point) 'face
667 `,(shr-face-background face))))
668 (setq start (point))
669 (shr-vertical-motion shr-internal-width)
670 (when (looking-at " $")
671 (delete-region (point) (line-end-position))))))
672
673 (defun shr-find-fill-point (start)
674 (let ((bp (point))
675 (end (point))
676 failed)
677 (while (not (or (setq failed (<= (point) start))
678 (eq (preceding-char) ? )
679 (eq (following-char) ? )
680 (shr-char-breakable-p (preceding-char))
681 (shr-char-breakable-p (following-char))
682 (and (shr-char-kinsoku-bol-p (preceding-char))
683 (shr-char-breakable-p (following-char))
684 (not (shr-char-kinsoku-bol-p (following-char))))
685 (shr-char-kinsoku-eol-p (following-char))
686 (bolp)))
687 (backward-char 1))
688 (if failed
689 ;; There's no breakable point, so we give it up.
690 (let (found)
691 (goto-char bp)
692 ;; Don't overflow the window edge, even if
693 ;; shr-kinsoku-shorten is nil.
694 (unless (or shr-kinsoku-shorten (null shr-width))
695 (while (setq found (re-search-forward
696 "\\(\\c>\\)\\| \\|\\c<\\|\\c|"
697 (line-end-position) 'move)))
698 (if (and found
699 (not (match-beginning 1)))
700 (goto-char (match-beginning 0)))))
701 (or
702 (eolp)
703 ;; Don't put kinsoku-bol characters at the beginning of a line,
704 ;; or kinsoku-eol characters at the end of a line.
705 (cond
706 ;; Don't overflow the window edge, even if shr-kinsoku-shorten
707 ;; is nil.
708 ((or shr-kinsoku-shorten (null shr-width))
709 (while (and (not (memq (preceding-char) (list ?\C-@ ?\n ? )))
710 (or (shr-char-kinsoku-eol-p (preceding-char))
711 (shr-char-kinsoku-bol-p (following-char))))
712 (backward-char 1))
713 (when (setq failed (<= (point) start))
714 ;; There's no breakable point that doesn't violate kinsoku,
715 ;; so we look for the second best position.
716 (while (and (progn
717 (forward-char 1)
718 (<= (point) end))
719 (progn
720 (setq bp (point))
721 (shr-char-kinsoku-eol-p (following-char)))))
722 (goto-char bp)))
723 ((shr-char-kinsoku-eol-p (preceding-char))
724 ;; Find backward the point where kinsoku-eol characters begin.
725 (let ((count 4))
726 (while
727 (progn
728 (backward-char 1)
729 (and (> (setq count (1- count)) 0)
730 (not (memq (preceding-char) (list ?\C-@ ?\n ? )))
731 (or (shr-char-kinsoku-eol-p (preceding-char))
732 (shr-char-kinsoku-bol-p (following-char)))))))
733 (when (setq failed (<= (point) start))
734 ;; There's no breakable point that doesn't violate kinsoku,
735 ;; so we go to the second best position.
736 (if (looking-at "\\(\\c<+\\)\\c<")
737 (goto-char (match-end 1))
738 (forward-char 1))))
739 ((shr-char-kinsoku-bol-p (following-char))
740 ;; Find forward the point where kinsoku-bol characters end.
741 (let ((count 4))
742 (while (progn
743 (forward-char 1)
744 (and (>= (setq count (1- count)) 0)
745 (shr-char-kinsoku-bol-p (following-char))
746 (shr-char-breakable-p (following-char))))))))
747 (when (eq (following-char) ? )
748 (forward-char 1))))
749 (not failed)))
750
751 (defun shr-parse-base (url)
752 ;; Always chop off anchors.
753 (when (string-match "#.*" url)
754 (setq url (substring url 0 (match-beginning 0))))
755 ;; NB: <base href="" > URI may itself be relative to the document s URI
756 (setq url (shr-expand-url url))
757 (let* ((parsed (url-generic-parse-url url))
758 (local (url-filename parsed)))
759 (setf (url-filename parsed) "")
760 ;; Chop off the bit after the last slash.
761 (when (string-match "\\`\\(.*/\\)[^/]+\\'" local)
762 (setq local (match-string 1 local)))
763 ;; Always make the local bit end with a slash.
764 (when (and (not (zerop (length local)))
765 (not (eq (aref local (1- (length local))) ?/)))
766 (setq local (concat local "/")))
767 (list (url-recreate-url parsed)
768 local
769 (url-type parsed)
770 url)))
771
772 (autoload 'url-expand-file-name "url-expand")
773
774 ;; FIXME This needs some tests writing.
775 ;; Does it even need to exist, given that url-expand-file-name does?
776 (defun shr-expand-url (url &optional base)
777 (setq base
778 (if base
779 ;; shr-parse-base should never call this with non-nil base!
780 (shr-parse-base base)
781 ;; Bound by the parser.
782 shr-base))
783 (when (zerop (length url))
784 (setq url nil))
785 ;; Strip leading whitespace
786 (and url (string-match "\\`\\s-+" url)
787 (setq url (substring url (match-end 0))))
788 (cond ((or (not url)
789 (not base)
790 (string-match "\\`[a-z]*:" url))
791 ;; Absolute or empty URI
792 (or url (nth 3 base)))
793 ((eq (aref url 0) ?/)
794 (if (and (> (length url) 1)
795 (eq (aref url 1) ?/))
796 ;; //host...; just use the protocol
797 (concat (nth 2 base) ":" url)
798 ;; Just use the host name part.
799 (concat (car base) url)))
800 ((eq (aref url 0) ?#)
801 ;; A link to an anchor.
802 (concat (nth 3 base) url))
803 (t
804 ;; Totally relative.
805 (url-expand-file-name url (concat (car base) (cadr base))))))
806
807 (defun shr-ensure-newline ()
808 (unless (bobp)
809 (let ((prefix (get-text-property (line-beginning-position)
810 'shr-prefix-length)))
811 (unless (or (zerop (current-column))
812 (and prefix
813 (= prefix (- (point) (line-beginning-position)))))
814 (insert "\n")))))
815
816 (defun shr-ensure-paragraph ()
817 (unless (bobp)
818 (let ((prefix (get-text-property (line-beginning-position)
819 'shr-prefix-length)))
820 (cond
821 ((and (bolp)
822 (save-excursion
823 (forward-line -1)
824 (looking-at " *$")))
825 ;; We're already at a new paragraph; do nothing.
826 )
827 ((and prefix
828 (= prefix (- (point) (line-beginning-position))))
829 ;; Do nothing; we're at the start of a <li>.
830 )
831 ((save-excursion
832 (beginning-of-line)
833 ;; If the current line is totally blank, and doesn't even
834 ;; have any face properties set, then delete the blank
835 ;; space.
836 (and (looking-at " *$")
837 (not (get-text-property (point) 'face))
838 (not (= (next-single-property-change (point) 'face nil
839 (line-end-position))
840 (line-end-position)))))
841 (delete-region (match-beginning 0) (match-end 0)))
842 ;; We have a single blank line.
843 ((and (eolp) (bolp))
844 (insert "\n"))
845 ;; Insert new paragraph.
846 (t
847 (insert "\n\n"))))))
848
849 (defun shr-indent ()
850 (when (> shr-indentation 0)
851 (insert
852 (if (not shr-use-fonts)
853 (make-string shr-indentation ?\s)
854 (propertize " "
855 'display
856 `(space :width (,shr-indentation)))))))
857
858 (defun shr-fontize-dom (dom &rest types)
859 (let ((start (point)))
860 (shr-generic dom)
861 (dolist (type types)
862 (shr-add-font start (point) type))))
863
864 ;; Add face to the region, but avoid putting the font properties on
865 ;; blank text at the start of the line, and the newline at the end, to
866 ;; avoid ugliness.
867 (defun shr-add-font (start end type)
868 (save-excursion
869 (goto-char start)
870 (while (< (point) end)
871 (when (bolp)
872 (skip-chars-forward " "))
873 (add-face-text-property (point) (min (line-end-position) end) type t)
874 (if (< (line-end-position) end)
875 (forward-line 1)
876 (goto-char end)))))
877
878 (defun shr-mouse-browse-url (ev)
879 "Browse the URL under the mouse cursor."
880 (interactive "e")
881 (mouse-set-point ev)
882 (shr-browse-url))
883
884 (defun shr-browse-url (&optional external mouse-event)
885 "Browse the URL under point.
886 If EXTERNAL, browse the URL using `shr-external-browser'."
887 (interactive (list current-prefix-arg last-nonmenu-event))
888 (mouse-set-point mouse-event)
889 (let ((url (get-text-property (point) 'shr-url)))
890 (cond
891 ((not url)
892 (message "No link under point"))
893 ((string-match "^mailto:" url)
894 (browse-url-mail url))
895 (t
896 (if external
897 (funcall shr-external-browser url)
898 (browse-url url))))))
899
900 (defun shr-save-contents (directory)
901 "Save the contents from URL in a file."
902 (interactive "DSave contents of URL to directory: ")
903 (let ((url (get-text-property (point) 'shr-url)))
904 (if (not url)
905 (message "No link under point")
906 (url-retrieve (shr-encode-url url)
907 'shr-store-contents (list url directory)
908 nil t))))
909
910 (defun shr-store-contents (status url directory)
911 (unless (plist-get status :error)
912 (when (or (search-forward "\n\n" nil t)
913 (search-forward "\r\n\r\n" nil t))
914 (write-region (point) (point-max)
915 (expand-file-name (file-name-nondirectory url)
916 directory)))))
917
918 (defun shr-image-fetched (status buffer start end &optional flags)
919 (let ((image-buffer (current-buffer)))
920 (when (and (buffer-name buffer)
921 (not (plist-get status :error)))
922 (url-store-in-cache image-buffer)
923 (when (or (search-forward "\n\n" nil t)
924 (search-forward "\r\n\r\n" nil t))
925 (let ((data (shr-parse-image-data)))
926 (with-current-buffer buffer
927 (save-excursion
928 (let ((alt (buffer-substring start end))
929 (properties (text-properties-at start))
930 (inhibit-read-only t))
931 (delete-region start end)
932 (goto-char start)
933 (funcall shr-put-image-function data alt flags)
934 (while properties
935 (let ((type (pop properties))
936 (value (pop properties)))
937 (unless (memq type '(display image-size))
938 (put-text-property start (point) type value))))))))))
939 (kill-buffer image-buffer)))
940
941 (defun shr-image-from-data (data)
942 "Return an image from the data: URI content DATA."
943 (when (string-match
944 "\\(\\([^/;,]+\\(/[^;,]+\\)?\\)\\(;[^;,]+\\)*\\)?,\\(.*\\)"
945 data)
946 (let ((param (match-string 4 data))
947 (payload (url-unhex-string (match-string 5 data))))
948 (when (string-match "^.*\\(;[ \t]*base64\\)$" param)
949 (setq payload (base64-decode-string payload)))
950 payload)))
951
952 ;; Behind display-graphic-p test.
953 (declare-function image-size "image.c" (spec &optional pixels frame))
954 (declare-function image-animate "image" (image &optional index limit))
955
956 (defun shr-put-image (spec alt &optional flags)
957 "Insert image SPEC with a string ALT. Return image.
958 SPEC is either an image data blob, or a list where the first
959 element is the data blob and the second element is the content-type."
960 (if (display-graphic-p)
961 (let* ((size (cdr (assq 'size flags)))
962 (data (if (consp spec)
963 (car spec)
964 spec))
965 (content-type (and (consp spec)
966 (cadr spec)))
967 (start (point))
968 (image (cond
969 ((eq size 'original)
970 (create-image data nil t :ascent 100
971 :format content-type))
972 ((eq content-type 'image/svg+xml)
973 (create-image data 'svg t :ascent 100))
974 ((eq size 'full)
975 (ignore-errors
976 (shr-rescale-image data content-type
977 (plist-get flags :width)
978 (plist-get flags :height))))
979 (t
980 (ignore-errors
981 (shr-rescale-image data content-type
982 (plist-get flags :width)
983 (plist-get flags :height)))))))
984 (when image
985 ;; When inserting big-ish pictures, put them at the
986 ;; beginning of the line.
987 (when (and (> (current-column) 0)
988 (> (car (image-size image t)) 400))
989 (insert "\n"))
990 (if (eq size 'original)
991 (insert-sliced-image image (or alt "*") nil 20 1)
992 (insert-image image (or alt "*")))
993 (put-text-property start (point) 'image-size size)
994 (when (and shr-image-animate
995 (cond ((fboundp 'image-multi-frame-p)
996 ;; Only animate multi-frame things that specify a
997 ;; delay; eg animated gifs as opposed to
998 ;; multi-page tiffs. FIXME?
999 (cdr (image-multi-frame-p image)))
1000 ((fboundp 'image-animated-p)
1001 (image-animated-p image))))
1002 (image-animate image nil 60)))
1003 image)
1004 (insert (or alt ""))))
1005
1006 (defun shr-rescale-image (data content-type width height)
1007 "Rescale DATA, if too big, to fit the current buffer.
1008 WIDTH and HEIGHT are the sizes given in the HTML data, if any."
1009 (if (not (and (fboundp 'imagemagick-types)
1010 (get-buffer-window (current-buffer))))
1011 (create-image data nil t :ascent 100)
1012 (let* ((edges (window-inside-pixel-edges
1013 (get-buffer-window (current-buffer))))
1014 (max-width (truncate (* shr-max-image-proportion
1015 (- (nth 2 edges) (nth 0 edges)))))
1016 (max-height (truncate (* shr-max-image-proportion
1017 (- (nth 3 edges) (nth 1 edges))))))
1018 (when (or (and width
1019 (> width max-width))
1020 (and height
1021 (> height max-height)))
1022 (setq width nil
1023 height nil))
1024 (if (and width height)
1025 (create-image
1026 data 'imagemagick t
1027 :ascent 100
1028 :width width
1029 :height height
1030 :format content-type)
1031 (create-image
1032 data 'imagemagick t
1033 :ascent 100
1034 :max-width max-width
1035 :max-height max-height
1036 :format content-type)))))
1037
1038 ;; url-cache-extract autoloads url-cache.
1039 (declare-function url-cache-create-filename "url-cache" (url))
1040 (autoload 'mm-disable-multibyte "mm-util")
1041 (autoload 'browse-url-mail "browse-url")
1042
1043 (defun shr-get-image-data (url)
1044 "Get image data for URL.
1045 Return a string with image data."
1046 (with-temp-buffer
1047 (mm-disable-multibyte)
1048 (when (ignore-errors
1049 (url-cache-extract (url-cache-create-filename (shr-encode-url url)))
1050 t)
1051 (when (or (search-forward "\n\n" nil t)
1052 (search-forward "\r\n\r\n" nil t))
1053 (shr-parse-image-data)))))
1054
1055 (declare-function libxml-parse-xml-region "xml.c"
1056 (start end &optional base-url discard-comments))
1057
1058 (defun shr-parse-image-data ()
1059 (let ((data (buffer-substring (point) (point-max)))
1060 (content-type
1061 (save-excursion
1062 (save-restriction
1063 (narrow-to-region (point-min) (point))
1064 (let ((content-type (mail-fetch-field "content-type")))
1065 (and content-type
1066 ;; Remove any comments in the type string.
1067 (intern (replace-regexp-in-string ";.*" "" content-type)
1068 obarray)))))))
1069 ;; SVG images may contain references to further images that we may
1070 ;; want to block. So special-case these by parsing the XML data
1071 ;; and remove the blocked bits.
1072 (when (eq content-type 'image/svg+xml)
1073 (setq data
1074 (shr-dom-to-xml
1075 (libxml-parse-xml-region (point) (point-max)))))
1076 (list data content-type)))
1077
1078 (defun shr-image-displayer (content-function)
1079 "Return a function to display an image.
1080 CONTENT-FUNCTION is a function to retrieve an image for a cid url that
1081 is an argument. The function to be returned takes three arguments URL,
1082 START, and END. Note that START and END should be markers."
1083 `(lambda (url start end)
1084 (when url
1085 (if (string-match "\\`cid:" url)
1086 ,(when content-function
1087 `(let ((image (funcall ,content-function
1088 (substring url (match-end 0)))))
1089 (when image
1090 (goto-char start)
1091 (funcall shr-put-image-function
1092 image (buffer-substring start end))
1093 (delete-region (point) end))))
1094 (url-retrieve url 'shr-image-fetched
1095 (list (current-buffer) start end)
1096 t t)))))
1097
1098 (defun shr-heading (dom &rest types)
1099 (shr-ensure-paragraph)
1100 (apply #'shr-fontize-dom dom types)
1101 (shr-ensure-paragraph))
1102
1103 (defun shr-urlify (start url &optional title)
1104 (shr-add-font start (point) 'shr-link)
1105 (add-text-properties
1106 start (point)
1107 (list 'shr-url url
1108 'help-echo (let ((iri (or (ignore-errors
1109 (decode-coding-string
1110 (url-unhex-string url)
1111 'utf-8 t))
1112 url)))
1113 (if title (format "%s (%s)" iri title) iri))
1114 'follow-link t
1115 'mouse-face 'highlight))
1116 ;; Don't overwrite any keymaps that are already in the buffer (i.e.,
1117 ;; image keymaps).
1118 (while (and start
1119 (< start (point)))
1120 (let ((next (next-single-property-change start 'keymap nil (point))))
1121 (if (get-text-property start 'keymap)
1122 (setq start next)
1123 (put-text-property start (or next (point)) 'keymap shr-map)))))
1124
1125 (defun shr-encode-url (url)
1126 "Encode URL."
1127 (browse-url-url-encode-chars url "[)$ ]"))
1128
1129 (autoload 'shr-color-visible "shr-color")
1130 (autoload 'shr-color->hexadecimal "shr-color")
1131
1132 (defun shr-color-check (fg bg)
1133 "Check that FG is visible on BG.
1134 Returns (fg bg) with corrected values.
1135 Returns nil if the colors that would be used are the default
1136 ones, in case fg and bg are nil."
1137 (when (or fg bg)
1138 (let ((fixed (cond ((null fg) 'fg)
1139 ((null bg) 'bg))))
1140 ;; Convert colors to hexadecimal, or set them to default.
1141 (let ((fg (or (shr-color->hexadecimal fg)
1142 (frame-parameter nil 'foreground-color)))
1143 (bg (or (shr-color->hexadecimal bg)
1144 (frame-parameter nil 'background-color))))
1145 (cond ((eq fixed 'bg)
1146 ;; Only return the new fg
1147 (list nil (cadr (shr-color-visible bg fg t))))
1148 ((eq fixed 'fg)
1149 ;; Invert args and results and return only the new bg
1150 (list (cadr (shr-color-visible fg bg t)) nil))
1151 (t
1152 (shr-color-visible bg fg)))))))
1153
1154 (defun shr-colorize-region (start end fg &optional bg)
1155 (when (and shr-use-colors
1156 (or fg bg)
1157 (>= (display-color-cells) 88))
1158 (let ((new-colors (shr-color-check fg bg)))
1159 (when new-colors
1160 (when fg
1161 (add-face-text-property start end
1162 (list :foreground (cadr new-colors))
1163 t))
1164 (when bg
1165 (add-face-text-property start end
1166 (list :background (car new-colors))
1167 t)))
1168 new-colors)))
1169
1170 ;;; Tag-specific rendering rules.
1171
1172 (defun shr-tag-html (dom)
1173 (let ((dir (dom-attr dom 'dir)))
1174 (cond
1175 ((equal dir "ltr")
1176 (setq bidi-paragraph-direction 'left-to-right))
1177 ((equal dir "rtl")
1178 (setq bidi-paragraph-direction 'right-to-left))
1179 ((equal dir "auto")
1180 (setq bidi-paragraph-direction nil))))
1181 (shr-generic dom))
1182
1183 (defun shr-tag-body (dom)
1184 (let* ((start (point))
1185 (fgcolor (or (dom-attr dom 'fgcolor) (dom-attr dom 'text)))
1186 (bgcolor (dom-attr dom 'bgcolor))
1187 (shr-stylesheet (list (cons 'color fgcolor)
1188 (cons 'background-color bgcolor))))
1189 (shr-generic dom)
1190 (shr-colorize-region start (point) fgcolor bgcolor)))
1191
1192 (defun shr-tag-style (_dom)
1193 )
1194
1195 (defun shr-tag-script (_dom)
1196 )
1197
1198 (defun shr-tag-comment (_dom)
1199 )
1200
1201 (defun shr-dom-to-xml (dom)
1202 (with-temp-buffer
1203 (shr-dom-print dom)
1204 (buffer-string)))
1205
1206 (defun shr-dom-print (dom)
1207 "Convert DOM into a string containing the xml representation."
1208 (insert (format "<%s" (dom-tag dom)))
1209 (dolist (attr (dom-attributes dom))
1210 ;; Ignore attributes that start with a colon because they are
1211 ;; private elements.
1212 (unless (= (aref (format "%s" (car attr)) 0) ?:)
1213 (insert (format " %s=\"%s\"" (car attr) (cdr attr)))))
1214 (insert ">")
1215 (let (url)
1216 (dolist (elem (dom-children dom))
1217 (cond
1218 ((stringp elem)
1219 (insert elem))
1220 ((eq (dom-tag elem) 'comment)
1221 )
1222 ((or (not (eq (dom-tag elem) 'image))
1223 ;; Filter out blocked elements inside the SVG image.
1224 (not (setq url (dom-attr elem ':xlink:href)))
1225 (not shr-blocked-images)
1226 (not (string-match shr-blocked-images url)))
1227 (insert " ")
1228 (shr-dom-print elem)))))
1229 (insert (format "</%s>" (dom-tag dom))))
1230
1231 (defun shr-tag-svg (dom)
1232 (when (and (image-type-available-p 'svg)
1233 (not shr-inhibit-images)
1234 (dom-attr dom 'width)
1235 (dom-attr dom 'height))
1236 (funcall shr-put-image-function (list (shr-dom-to-xml dom) 'image/svg+xml)
1237 "SVG Image")))
1238
1239 (defun shr-tag-sup (dom)
1240 (let ((start (point)))
1241 (shr-generic dom)
1242 (put-text-property start (point) 'display '(raise 0.5))))
1243
1244 (defun shr-tag-sub (dom)
1245 (let ((start (point)))
1246 (shr-generic dom)
1247 (put-text-property start (point) 'display '(raise -0.5))))
1248
1249 (defun shr-tag-label (dom)
1250 (shr-generic dom)
1251 (shr-ensure-paragraph))
1252
1253 (defun shr-tag-p (dom)
1254 (shr-ensure-paragraph)
1255 (shr-generic dom)
1256 (shr-ensure-paragraph))
1257
1258 (defun shr-tag-div (dom)
1259 (shr-ensure-newline)
1260 (shr-generic dom)
1261 (shr-ensure-newline))
1262
1263 (defun shr-tag-s (dom)
1264 (shr-fontize-dom dom 'shr-strike-through))
1265
1266 (defun shr-tag-b (dom)
1267 (shr-fontize-dom dom 'bold))
1268
1269 (defun shr-tag-i (dom)
1270 (shr-fontize-dom dom 'italic))
1271
1272 (defun shr-tag-em (dom)
1273 (shr-fontize-dom dom 'italic))
1274
1275 (defun shr-tag-strong (dom)
1276 (shr-fontize-dom dom 'bold))
1277
1278 (defun shr-tag-u (dom)
1279 (shr-fontize-dom dom 'underline))
1280
1281 (defun shr-tag-tt (dom)
1282 (let ((shr-current-font 'default))
1283 (shr-generic dom)))
1284
1285 (defun shr-tag-ins (cont)
1286 (let* ((start (point))
1287 (color "green")
1288 (shr-stylesheet (nconc (list (cons 'color color))
1289 shr-stylesheet)))
1290 (shr-generic cont)
1291 (shr-colorize-region start (point) color
1292 (cdr (assq 'background-color shr-stylesheet)))))
1293
1294 (defun shr-tag-del (cont)
1295 (let* ((start (point))
1296 (color "red")
1297 (shr-stylesheet (nconc (list (cons 'color color))
1298 shr-stylesheet)))
1299 (shr-fontize-dom cont 'shr-strike-through)
1300 (shr-colorize-region start (point) color
1301 (cdr (assq 'background-color shr-stylesheet)))))
1302
1303 (defun shr-parse-style (style)
1304 (when style
1305 (save-match-data
1306 (when (string-match "\n" style)
1307 (setq style (replace-match " " t t style))))
1308 (let ((plist nil))
1309 (dolist (elem (split-string style ";"))
1310 (when elem
1311 (setq elem (split-string elem ":"))
1312 (when (and (car elem)
1313 (cadr elem))
1314 (let ((name (replace-regexp-in-string "^ +\\| +$" "" (car elem)))
1315 (value (replace-regexp-in-string "^ +\\| +$" "" (cadr elem))))
1316 (when (string-match " *!important\\'" value)
1317 (setq value (substring value 0 (match-beginning 0))))
1318 (unless (equal value "inherit")
1319 (push (cons (intern name obarray)
1320 value)
1321 plist))))))
1322 plist)))
1323
1324 (defun shr-tag-base (dom)
1325 (when-let (base (dom-attr dom 'href))
1326 (setq shr-base (shr-parse-base base)))
1327 (shr-generic dom))
1328
1329 (defun shr-tag-a (dom)
1330 (let ((url (dom-attr dom 'href))
1331 (title (dom-attr dom 'title))
1332 (start (point))
1333 shr-start)
1334 (shr-generic dom)
1335 (when (and shr-target-id
1336 (equal (dom-attr dom 'name) shr-target-id))
1337 ;; We have a zero-length <a name="foo"> element, so just
1338 ;; insert... something.
1339 (when (= start (point))
1340 (shr-ensure-newline)
1341 (insert " "))
1342 (put-text-property start (1+ start) 'shr-target-id shr-target-id))
1343 (when url
1344 (shr-urlify (or shr-start start) (shr-expand-url url) title))))
1345
1346 (defun shr-tag-object (dom)
1347 (unless shr-inhibit-images
1348 (let ((start (point))
1349 url multimedia image)
1350 (when-let (type (dom-attr dom 'type))
1351 (when (string-match "\\`image/svg" type)
1352 (setq url (dom-attr dom 'data)
1353 image t)))
1354 (dolist (child (dom-non-text-children dom))
1355 (cond
1356 ((eq (dom-tag child) 'embed)
1357 (setq url (or url (dom-attr child 'src))
1358 multimedia t))
1359 ((and (eq (dom-tag child) 'param)
1360 (equal (dom-attr child 'name) "movie"))
1361 (setq url (or url (dom-attr child 'value))
1362 multimedia t))))
1363 (when url
1364 (cond
1365 (image
1366 (shr-tag-img dom url)
1367 (setq dom nil))
1368 (multimedia
1369 (shr-insert " [multimedia] ")
1370 (shr-urlify start (shr-expand-url url)))))
1371 (when dom
1372 (shr-generic dom)))))
1373
1374 (defcustom shr-prefer-media-type-alist '(("webm" . 1.0)
1375 ("ogv" . 1.0)
1376 ("ogg" . 1.0)
1377 ("opus" . 1.0)
1378 ("flac" . 0.9)
1379 ("wav" . 0.5))
1380 "Preferences for media types.
1381 The key element should be a regexp matched against the type of the source or
1382 url if no type is specified. The value should be a float in the range 0.0 to
1383 1.0. Media elements with higher value are preferred."
1384 :version "24.4"
1385 :group 'shr
1386 :type '(alist :key-type regexp :value-type float))
1387
1388 (defun shr--get-media-pref (elem)
1389 "Determine the preference for ELEM.
1390 The preference is a float determined from `shr-prefer-media-type'."
1391 (let ((type (dom-attr elem 'type))
1392 (p 0.0))
1393 (unless type
1394 (setq type (dom-attr elem 'src)))
1395 (when type
1396 (dolist (pref shr-prefer-media-type-alist)
1397 (when (and
1398 (> (cdr pref) p)
1399 (string-match-p (car pref) type))
1400 (setq p (cdr pref)))))
1401 p))
1402
1403 (defun shr--extract-best-source (dom &optional url pref)
1404 "Extract the best `:src' property from <source> blocks in DOM."
1405 (setq pref (or pref -1.0))
1406 (let (new-pref)
1407 (dolist (elem (dom-non-text-children dom))
1408 (when (and (eq (dom-tag elem) 'source)
1409 (< pref
1410 (setq new-pref
1411 (shr--get-media-pref elem))))
1412 (setq pref new-pref
1413 url (dom-attr elem 'src))
1414 ;; libxml's html parser isn't HTML5 compliant and non terminated
1415 ;; source tags might end up as children. So recursion it is...
1416 (dolist (child (dom-non-text-children elem))
1417 (when (eq (dom-tag child) 'source)
1418 (let ((ret (shr--extract-best-source (list child) url pref)))
1419 (when (< pref (cdr ret))
1420 (setq url (car ret)
1421 pref (cdr ret)))))))))
1422 (cons url pref))
1423
1424 (defun shr-tag-video (dom)
1425 (let ((image (dom-attr dom 'poster))
1426 (url (dom-attr dom 'src))
1427 (start (point)))
1428 (unless url
1429 (setq url (car (shr--extract-best-source dom))))
1430 (if (> (length image) 0)
1431 (shr-tag-img nil image)
1432 (shr-insert " [video] "))
1433 (shr-urlify start (shr-expand-url url))))
1434
1435 (defun shr-tag-audio (dom)
1436 (let ((url (dom-attr dom 'src))
1437 (start (point)))
1438 (unless url
1439 (setq url (car (shr--extract-best-source dom))))
1440 (shr-insert " [audio] ")
1441 (shr-urlify start (shr-expand-url url))))
1442
1443 (defun shr-tag-img (dom &optional url)
1444 (when (or url
1445 (and dom
1446 (> (length (dom-attr dom 'src)) 0)))
1447 (when (> (current-column) 0)
1448 (insert "\n"))
1449 (let ((alt (dom-attr dom 'alt))
1450 (width (shr-string-number (dom-attr dom 'width)))
1451 (height (shr-string-number (dom-attr dom 'height)))
1452 (url (shr-expand-url (or url (dom-attr dom 'src)))))
1453 (let ((start (point-marker)))
1454 (when (zerop (length alt))
1455 (setq alt "*"))
1456 (cond
1457 ((or (member (dom-attr dom 'height) '("0" "1"))
1458 (member (dom-attr dom 'width) '("0" "1")))
1459 ;; Ignore zero-sized or single-pixel images.
1460 )
1461 ((and (not shr-inhibit-images)
1462 (string-match "\\`data:" url))
1463 (let ((image (shr-image-from-data (substring url (match-end 0)))))
1464 (if image
1465 (funcall shr-put-image-function image alt
1466 (list :width width :height height))
1467 (insert alt))))
1468 ((and (not shr-inhibit-images)
1469 (string-match "\\`cid:" url))
1470 (let ((url (substring url (match-end 0)))
1471 image)
1472 (if (or (not shr-content-function)
1473 (not (setq image (funcall shr-content-function url))))
1474 (insert alt)
1475 (funcall shr-put-image-function image alt
1476 (list :width width :height height)))))
1477 ((or shr-inhibit-images
1478 (and shr-blocked-images
1479 (string-match shr-blocked-images url)))
1480 (setq shr-start (point))
1481 (shr-insert alt))
1482 ((and (not shr-ignore-cache)
1483 (url-is-cached (shr-encode-url url)))
1484 (funcall shr-put-image-function (shr-get-image-data url) alt
1485 (list :width width :height height)))
1486 (t
1487 (when (and shr-ignore-cache
1488 (url-is-cached (shr-encode-url url)))
1489 (let ((file (url-cache-create-filename (shr-encode-url url))))
1490 (when (file-exists-p file)
1491 (delete-file file))))
1492 (when (image-type-available-p 'svg)
1493 (insert-image
1494 (shr-make-placeholder-image dom)
1495 (or alt "")))
1496 (insert " ")
1497 (url-queue-retrieve
1498 (shr-encode-url url) 'shr-image-fetched
1499 (list (current-buffer) start (set-marker (make-marker) (point))
1500 (list :width width :height height))
1501 t t)))
1502 (when (zerop shr-table-depth) ;; We are not in a table.
1503 (put-text-property start (point) 'keymap shr-image-map)
1504 (put-text-property start (point) 'shr-alt alt)
1505 (put-text-property start (point) 'image-url url)
1506 (put-text-property start (point) 'image-displayer
1507 (shr-image-displayer shr-content-function))
1508 (put-text-property start (point) 'help-echo
1509 (shr-fill-text
1510 (or (dom-attr dom 'title) alt))))))))
1511
1512 (defun shr-string-number (string)
1513 (if (null string)
1514 nil
1515 (setq string (replace-regexp-in-string "[^0-9]" "" string))
1516 (if (zerop (length string))
1517 nil
1518 (string-to-number string))))
1519
1520 (defun shr-make-placeholder-image (dom)
1521 (let* ((edges (and
1522 (get-buffer-window (current-buffer))
1523 (window-inside-pixel-edges
1524 (get-buffer-window (current-buffer)))))
1525 (scaling (image-compute-scaling-factor image-scaling-factor))
1526 (width (truncate
1527 (* (or (shr-string-number (dom-attr dom 'width)) 100)
1528 scaling)))
1529 (height (truncate
1530 (* (or (shr-string-number (dom-attr dom 'height)) 100)
1531 scaling)))
1532 (max-width
1533 (and edges
1534 (truncate (* shr-max-image-proportion
1535 (- (nth 2 edges) (nth 0 edges))))))
1536 (max-height (and edges
1537 (truncate (* shr-max-image-proportion
1538 (- (nth 3 edges) (nth 1 edges))))))
1539 svg image)
1540 (when (and max-width
1541 (> width max-width))
1542 (setq height (truncate (* (/ (float max-width) width) height))
1543 width max-width))
1544 (when (and max-height
1545 (> height max-height))
1546 (setq width (truncate (* (/ (float max-height) height) width))
1547 height max-height))
1548 (setq svg (svg-create width height))
1549 (svg-gradient svg "background" 'linear '((0 . "#b0b0b0") (100 . "#808080")))
1550 (svg-rectangle svg 0 0 width height :gradient "background"
1551 :stroke-width 2 :stroke-color "black")
1552 (let ((image (svg-image svg)))
1553 (setf (image-property image :ascent) 100)
1554 image)))
1555
1556 (defun shr-tag-pre (dom)
1557 (let ((shr-folding-mode 'none)
1558 (shr-current-font 'default))
1559 (shr-ensure-newline)
1560 (shr-generic dom)
1561 (shr-ensure-newline)))
1562
1563 (defun shr-tag-blockquote (dom)
1564 (shr-ensure-paragraph)
1565 (let ((start (point))
1566 (shr-indentation (+ shr-indentation
1567 (* 4 shr-table-separator-pixel-width))))
1568 (shr-generic dom)
1569 (shr-ensure-paragraph)
1570 (shr-mark-fill start)))
1571
1572 (defun shr-tag-dl (dom)
1573 (shr-ensure-paragraph)
1574 (shr-generic dom)
1575 (shr-ensure-paragraph))
1576
1577 (defun shr-tag-dt (dom)
1578 (shr-ensure-newline)
1579 (shr-generic dom)
1580 (shr-ensure-newline))
1581
1582 (defun shr-tag-dd (dom)
1583 (shr-ensure-newline)
1584 (let ((shr-indentation (+ shr-indentation
1585 (* 4 shr-table-separator-pixel-width))))
1586 (shr-generic dom)))
1587
1588 (defun shr-tag-ul (dom)
1589 (shr-ensure-paragraph)
1590 (let ((shr-list-mode 'ul))
1591 (shr-generic dom))
1592 ;; If we end on an empty <li>, then make sure we really end on a new
1593 ;; paragraph.
1594 (unless (bolp)
1595 (insert "\n"))
1596 (shr-ensure-paragraph))
1597
1598 (defun shr-tag-ol (dom)
1599 (shr-ensure-paragraph)
1600 (let ((shr-list-mode 1))
1601 (shr-generic dom))
1602 (shr-ensure-paragraph))
1603
1604 (defun shr-tag-li (dom)
1605 (shr-ensure-newline)
1606 (let ((start (point)))
1607 (let* ((bullet
1608 (if (numberp shr-list-mode)
1609 (prog1
1610 (format "%d " shr-list-mode)
1611 (setq shr-list-mode (1+ shr-list-mode)))
1612 (car shr-internal-bullet)))
1613 (width (if (numberp shr-list-mode)
1614 (shr-string-pixel-width bullet)
1615 (cdr shr-internal-bullet))))
1616 (insert bullet)
1617 (shr-mark-fill start)
1618 (let ((shr-indentation (+ shr-indentation width)))
1619 (put-text-property start (1+ start)
1620 'shr-continuation-indentation shr-indentation)
1621 (put-text-property start (1+ start) 'shr-prefix-length (length bullet))
1622 (shr-generic dom)))))
1623
1624 (defun shr-mark-fill (start)
1625 ;; We may not have inserted any text to fill.
1626 (unless (= start (point))
1627 (put-text-property start (1+ start)
1628 'shr-indentation shr-indentation)))
1629
1630 (defun shr-tag-br (dom)
1631 (when (and (not (bobp))
1632 ;; Only add a newline if we break the current line, or
1633 ;; the previous line isn't a blank line.
1634 (or (not (bolp))
1635 (and (> (- (point) 2) (point-min))
1636 (not (= (char-after (- (point) 2)) ?\n)))))
1637 (insert "\n"))
1638 (shr-generic dom))
1639
1640 (defun shr-tag-span (dom)
1641 (shr-generic dom))
1642
1643 (defun shr-tag-h1 (dom)
1644 (shr-heading dom (if shr-use-fonts
1645 '(variable-pitch (:height 1.3 :weight bold))
1646 'bold)))
1647
1648 (defun shr-tag-h2 (dom)
1649 (shr-heading dom 'bold))
1650
1651 (defun shr-tag-h3 (dom)
1652 (shr-heading dom 'italic))
1653
1654 (defun shr-tag-h4 (dom)
1655 (shr-heading dom))
1656
1657 (defun shr-tag-h5 (dom)
1658 (shr-heading dom))
1659
1660 (defun shr-tag-h6 (dom)
1661 (shr-heading dom))
1662
1663 (defun shr-tag-hr (_dom)
1664 (shr-ensure-newline)
1665 (insert (make-string (if (not shr-use-fonts)
1666 shr-internal-width
1667 (1+ (/ shr-internal-width
1668 shr-table-separator-pixel-width)))
1669 shr-hr-line)
1670 "\n"))
1671
1672 (defun shr-tag-title (dom)
1673 (shr-heading dom 'bold 'underline))
1674
1675 (defun shr-tag-font (dom)
1676 (let* ((start (point))
1677 (color (dom-attr dom 'color))
1678 (shr-stylesheet (nconc (list (cons 'color color))
1679 shr-stylesheet)))
1680 (shr-generic dom)
1681 (when color
1682 (shr-colorize-region start (point) color
1683 (cdr (assq 'background-color shr-stylesheet))))))
1684
1685 (defun shr-tag-bdo (dom)
1686 (let* ((direction (dom-attr dom 'dir))
1687 (char (cond
1688 ((equal direction "ltr")
1689 #x202d) ; LRO
1690 ((equal direction "rtl")
1691 #x202e)))) ; RLO
1692 (when char
1693 (insert #x2068 char)) ; FSI + LRO/RLO
1694 (shr-generic dom)
1695 (when char
1696 (insert #x202c #x2069)))) ; PDF + PDI
1697
1698 (defun shr-tag-bdi (dom)
1699 (insert #x2068) ; FSI
1700 (shr-generic dom)
1701 (insert #x2069)) ; PDI
1702
1703 ;;; Table rendering algorithm.
1704
1705 ;; Table rendering is the only complicated thing here. We do this by
1706 ;; first counting how many TDs there are in each TR, and registering
1707 ;; how wide they think they should be ("width=45%", etc). Then we
1708 ;; render each TD separately (this is done in temporary buffers, so
1709 ;; that we can use all the rendering machinery as if we were in the
1710 ;; main buffer). Now we know how much space each TD really takes, so
1711 ;; we then render everything again with the new widths, and finally
1712 ;; insert all these boxes into the main buffer.
1713 (defun shr-tag-table-1 (dom)
1714 (setq dom (or (dom-child-by-tag dom 'tbody) dom))
1715 (let* ((shr-inhibit-images t)
1716 (shr-table-depth (1+ shr-table-depth))
1717 (shr-kinsoku-shorten t)
1718 ;; Find all suggested widths.
1719 (columns (shr-column-specs dom))
1720 ;; Compute how many pixels wide each TD should be.
1721 (suggested-widths (shr-pro-rate-columns columns))
1722 ;; Do a "test rendering" to see how big each TD is (this can
1723 ;; be smaller (if there's little text) or bigger (if there's
1724 ;; unbreakable text).
1725 (elems (or (dom-attr dom 'shr-suggested-widths)
1726 (shr-make-table dom suggested-widths nil
1727 'shr-suggested-widths)))
1728 (sketch (loop for line in elems
1729 collect (mapcar #'car line)))
1730 (natural (loop for line in elems
1731 collect (mapcar #'cdr line)))
1732 (sketch-widths (shr-table-widths sketch natural suggested-widths)))
1733 ;; This probably won't work very well.
1734 (when (> (+ (loop for width across sketch-widths
1735 summing (1+ width))
1736 shr-indentation shr-table-separator-pixel-width)
1737 (frame-width))
1738 (setq truncate-lines t))
1739 ;; Then render the table again with these new "hard" widths.
1740 (shr-insert-table (shr-make-table dom sketch-widths t) sketch-widths)))
1741
1742 (defun shr-table-body (dom)
1743 (let ((tbodies (seq-filter (lambda (child)
1744 (eq (dom-tag child) 'tbody))
1745 (dom-non-text-children dom))))
1746 (cond
1747 ((null tbodies)
1748 dom)
1749 ((= (length tbodies) 1)
1750 (car tbodies))
1751 (t
1752 ;; Table with multiple tbodies. Convert into a single tbody.
1753 `(tbody nil ,@(cl-reduce 'append
1754 (mapcar 'dom-non-text-children tbodies)))))))
1755
1756 (defun shr-tag-table (dom)
1757 (shr-ensure-paragraph)
1758 (let* ((caption (dom-children (dom-child-by-tag dom 'caption)))
1759 (header (dom-non-text-children (dom-child-by-tag dom 'thead)))
1760 (body (dom-non-text-children (shr-table-body dom)))
1761 (footer (dom-non-text-children (dom-child-by-tag dom 'tfoot)))
1762 (bgcolor (dom-attr dom 'bgcolor))
1763 (start (point))
1764 (shr-stylesheet (nconc (list (cons 'background-color bgcolor))
1765 shr-stylesheet))
1766 (nheader (if header (shr-max-columns header)))
1767 (nbody (if body (shr-max-columns body) 0))
1768 (nfooter (if footer (shr-max-columns footer))))
1769 (if (and (not caption)
1770 (not header)
1771 (not (dom-child-by-tag dom 'tbody))
1772 (not (dom-child-by-tag dom 'tr))
1773 (not footer))
1774 ;; The table is totally invalid and just contains random junk.
1775 ;; Try to output it anyway.
1776 (shr-generic dom)
1777 ;; It's a real table, so render it.
1778 (if (dom-attr dom 'shr-fixed-table)
1779 (shr-tag-table-1 dom)
1780 ;; Only fix up the table once.
1781 (let ((table
1782 (nconc
1783 (list 'table nil)
1784 (if caption `((tr nil (td nil ,@caption))))
1785 (cond
1786 (header
1787 (if footer
1788 ;; header + body + footer
1789 (if (= nheader nbody)
1790 (if (= nbody nfooter)
1791 `((tr nil (td nil (table nil
1792 (tbody nil ,@header
1793 ,@body ,@footer)))))
1794 (nconc `((tr nil (td nil (table nil
1795 (tbody nil ,@header
1796 ,@body)))))
1797 (if (= nfooter 1)
1798 footer
1799 `((tr nil (td nil (table
1800 nil (tbody
1801 nil ,@footer))))))))
1802 (nconc `((tr nil (td nil (table nil (tbody
1803 nil ,@header)))))
1804 (if (= nbody nfooter)
1805 `((tr nil (td nil (table
1806 nil (tbody nil ,@body
1807 ,@footer)))))
1808 (nconc `((tr nil (td nil (table
1809 nil (tbody nil
1810 ,@body)))))
1811 (if (= nfooter 1)
1812 footer
1813 `((tr nil (td nil (table
1814 nil
1815 (tbody
1816 nil
1817 ,@footer))))))))))
1818 ;; header + body
1819 (if (= nheader nbody)
1820 `((tr nil (td nil (table nil (tbody nil ,@header
1821 ,@body)))))
1822 (if (= nheader 1)
1823 `(,@header (tr nil (td nil (table
1824 nil (tbody nil ,@body)))))
1825 `((tr nil (td nil (table nil (tbody nil ,@header))))
1826 (tr nil (td nil (table nil (tbody nil ,@body)))))))))
1827 (footer
1828 ;; body + footer
1829 (if (= nbody nfooter)
1830 `((tr nil (td nil (table
1831 nil (tbody nil ,@body ,@footer)))))
1832 (nconc `((tr nil (td nil (table nil (tbody nil ,@body)))))
1833 (if (= nfooter 1)
1834 footer
1835 `((tr nil (td nil (table
1836 nil (tbody nil ,@footer)))))))))
1837 (caption
1838 `((tr nil (td nil (table nil (tbody nil ,@body))))))
1839 (body)))))
1840 (dom-set-attribute table 'shr-fixed-table t)
1841 (setcdr dom (cdr table))
1842 (shr-tag-table-1 dom))))
1843 (when bgcolor
1844 (shr-colorize-region start (point) (cdr (assq 'color shr-stylesheet))
1845 bgcolor))
1846 ;; Finally, insert all the images after the table. The Emacs buffer
1847 ;; model isn't strong enough to allow us to put the images actually
1848 ;; into the tables.
1849 (when (zerop shr-table-depth)
1850 (save-excursion
1851 (shr-expand-alignments start (point)))
1852 (dolist (elem (dom-by-tag dom 'object))
1853 (shr-tag-object elem))
1854 (dolist (elem (dom-by-tag dom 'img))
1855 (shr-tag-img elem)))))
1856
1857 (defun shr-insert-table (table widths)
1858 (let* ((collapse (equal (cdr (assq 'border-collapse shr-stylesheet))
1859 "collapse"))
1860 (shr-table-separator-length (if collapse 0 1))
1861 (shr-table-vertical-line (if collapse "" shr-table-vertical-line))
1862 (start (point)))
1863 (setq shr-table-id (1+ shr-table-id))
1864 (unless collapse
1865 (shr-insert-table-ruler widths))
1866 (dolist (row table)
1867 (let ((start (point))
1868 (align 0)
1869 (column-number 0)
1870 (height (let ((max 0))
1871 (dolist (column row)
1872 (setq max (max max (nth 2 column))))
1873 max)))
1874 (dotimes (i (max height 1))
1875 (shr-indent)
1876 (insert shr-table-vertical-line "\n"))
1877 (dolist (column row)
1878 (when (> (nth 2 column) -1)
1879 (goto-char start)
1880 ;; Sum up all the widths from the column. (There may be
1881 ;; more than one if this is a "colspan" column.)
1882 (dotimes (i (nth 4 column))
1883 ;; The colspan directive may be wrong and there may not be
1884 ;; that number of columns.
1885 (when (<= column-number (1- (length widths)))
1886 (setq align (+ align
1887 (aref widths column-number)
1888 (* 2 shr-table-separator-pixel-width))))
1889 (setq column-number (1+ column-number)))
1890 (let ((lines (nth 3 column))
1891 (pixel-align (if (not shr-use-fonts)
1892 (* align (frame-char-width))
1893 align)))
1894 (dolist (line lines)
1895 (end-of-line)
1896 (let ((start (point))
1897 (background (and (> (length line) 0)
1898 (shr-face-background
1899 (get-text-property
1900 (1- (length line)) 'face line))))
1901 (space (propertize
1902 " "
1903 'display `(space :align-to (,pixel-align))
1904 'shr-table-indent shr-table-id)))
1905 (when background
1906 (setq space (propertize space 'face background)))
1907 (insert line space shr-table-vertical-line)
1908 (shr-colorize-region
1909 start (1- (point)) (nth 5 column) (nth 6 column)))
1910 (forward-line 1))
1911 ;; Add blank lines at padding at the bottom of the TD,
1912 ;; possibly.
1913 (dotimes (i (- height (length lines)))
1914 (end-of-line)
1915 (let ((start (point)))
1916 (insert (propertize " "
1917 'display `(space :align-to (,pixel-align))
1918 'shr-table-indent shr-table-id)
1919 shr-table-vertical-line)
1920 (shr-colorize-region
1921 start (1- (point)) (nth 5 column) (nth 6 column)))
1922 (forward-line 1))))))
1923 (unless collapse
1924 (shr-insert-table-ruler widths)))
1925 (unless (= start (point))
1926 (put-text-property start (1+ start) 'shr-table-id shr-table-id))))
1927
1928 (defun shr-face-background (face)
1929 (and (consp face)
1930 (or (and (plist-get face :background)
1931 (list :background (plist-get face :background)))
1932 (let ((background nil))
1933 (dolist (elem face)
1934 (when (and (consp elem)
1935 (eq (car elem) :background)
1936 (not background))
1937 (setq background (cadr elem))))
1938 (and background
1939 (list :background background))))))
1940
1941 (defun shr-expand-alignments (start end)
1942 (while (< (setq start (next-single-property-change
1943 start 'shr-table-id nil end))
1944 end)
1945 (goto-char start)
1946 (let* ((shr-use-fonts t)
1947 (id (get-text-property (point) 'shr-table-id))
1948 (base (shr-pixel-column))
1949 elem)
1950 (when id
1951 (save-excursion
1952 (while (setq elem (text-property-any
1953 (point) end 'shr-table-indent id))
1954 (goto-char elem)
1955 (let ((align (get-text-property (point) 'display)))
1956 (put-text-property (point) (1+ (point)) 'display
1957 `(space :align-to (,(+ (car (nth 2 align))
1958 base)))))
1959 (forward-char 1)))))
1960 (setq start (1+ start))))
1961
1962 (defun shr-insert-table-ruler (widths)
1963 (when shr-table-horizontal-line
1964 (when (and (bolp)
1965 (> shr-indentation 0))
1966 (shr-indent))
1967 (insert shr-table-corner)
1968 (let ((total-width 0))
1969 (dotimes (i (length widths))
1970 (setq total-width (+ total-width (aref widths i)
1971 (* shr-table-separator-pixel-width 2)))
1972 (insert (make-string (1+ (/ (aref widths i)
1973 shr-table-separator-pixel-width))
1974 shr-table-horizontal-line)
1975 (propertize " "
1976 'display `(space :align-to (,total-width))
1977 'shr-table-indent shr-table-id)
1978 shr-table-corner)))
1979 (insert "\n")))
1980
1981 (defun shr-table-widths (table natural-table suggested-widths)
1982 (let* ((length (length suggested-widths))
1983 (widths (make-vector length 0))
1984 (natural-widths (make-vector length 0)))
1985 (dolist (row table)
1986 (let ((i 0))
1987 (dolist (column row)
1988 (aset widths i (max (aref widths i) column))
1989 (setq i (1+ i)))))
1990 (dolist (row natural-table)
1991 (let ((i 0))
1992 (dolist (column row)
1993 (aset natural-widths i (max (aref natural-widths i) column))
1994 (setq i (1+ i)))))
1995 (let ((extra (- (apply '+ (append suggested-widths nil))
1996 (apply '+ (append widths nil))
1997 (* shr-table-separator-pixel-width (1+ (length widths)))))
1998 (expanded-columns 0))
1999 ;; We have extra, unused space, so divide this space amongst the
2000 ;; columns.
2001 (when (> extra 0)
2002 ;; If the natural width is wider than the rendered width, we
2003 ;; want to allow the column to expand.
2004 (dotimes (i length)
2005 (when (> (aref natural-widths i) (aref widths i))
2006 (setq expanded-columns (1+ expanded-columns))))
2007 (dotimes (i length)
2008 (when (> (aref natural-widths i) (aref widths i))
2009 (aset widths i (min
2010 (aref natural-widths i)
2011 (+ (/ extra expanded-columns)
2012 (aref widths i))))))))
2013 widths))
2014
2015 (defun shr-make-table (dom widths &optional fill storage-attribute)
2016 (or (cadr (assoc (list dom widths fill) shr-content-cache))
2017 (let ((data (shr-make-table-1 dom widths fill)))
2018 (push (list (list dom widths fill) data)
2019 shr-content-cache)
2020 (when storage-attribute
2021 (dom-set-attribute dom storage-attribute data))
2022 data)))
2023
2024 (defun shr-make-table-1 (dom widths &optional fill)
2025 (let ((trs nil)
2026 (rowspans (make-vector (length widths) 0))
2027 (colspan-remaining 0)
2028 colspan-width colspan-count
2029 width colspan)
2030 (dolist (row (dom-non-text-children dom))
2031 (when (eq (dom-tag row) 'tr)
2032 (let ((tds nil)
2033 (columns (dom-non-text-children row))
2034 (i 0)
2035 (width-column 0)
2036 column)
2037 (while (< i (length widths))
2038 ;; If we previously had a rowspan definition, then that
2039 ;; means that we now have a "missing" td/th element here.
2040 ;; So just insert a dummy, empty one to (sort of) emulate
2041 ;; rowspan.
2042 (setq column
2043 (if (zerop (aref rowspans i))
2044 (pop columns)
2045 (aset rowspans i (1- (aref rowspans i)))
2046 '(td)))
2047 (when (and (not (stringp column))
2048 (or (memq (dom-tag column) '(td th))
2049 (not column)))
2050 (when-let (span (dom-attr column 'rowspan))
2051 (aset rowspans i (+ (aref rowspans i)
2052 (1- (string-to-number span)))))
2053 ;; Sanity check for invalid column-spans.
2054 (when (>= width-column (length widths))
2055 (setq width-column 0))
2056 (setq width
2057 (if column
2058 (aref widths width-column)
2059 (* 10 shr-table-separator-pixel-width)))
2060 (when (setq colspan (dom-attr column 'colspan))
2061 (setq colspan (min (string-to-number colspan)
2062 ;; The colspan may be wrong, so
2063 ;; truncate it to the length of the
2064 ;; remaining columns.
2065 (- (length widths) i)))
2066 (dotimes (j (1- colspan))
2067 (setq width
2068 (if (> (+ i 1 j) (1- (length widths)))
2069 ;; If we have a colspan spec that's longer
2070 ;; than the table is wide, just use the last
2071 ;; width as the width.
2072 (aref widths (1- (length widths)))
2073 ;; Sum up the widths of the columns we're
2074 ;; spanning.
2075 (+ width
2076 shr-table-separator-length
2077 (aref widths (+ i 1 j))))))
2078 (setq width-column (+ width-column (1- colspan))
2079 colspan-count colspan
2080 colspan-remaining colspan))
2081 (when column
2082 (let ((data (shr-render-td column width fill)))
2083 (if (and (not fill)
2084 (> colspan-remaining 0))
2085 (progn
2086 (setq colspan-width (car data))
2087 (let ((this-width (/ colspan-width colspan-count)))
2088 (push (cons this-width (cadr data)) tds)
2089 (setq colspan-remaining (1- colspan-remaining))))
2090 (if (not fill)
2091 (push (cons (car data) (cadr data)) tds)
2092 (push data tds)))))
2093 (when (and colspan
2094 (> colspan 1))
2095 (dotimes (c (1- colspan))
2096 (setq i (1+ i))
2097 (push
2098 (if fill
2099 (list 0 0 -1 nil 1 nil nil)
2100 '(0 . 0))
2101 tds)))
2102 (setq i (1+ i)
2103 width-column (1+ width-column))))
2104 (push (nreverse tds) trs))))
2105 (nreverse trs)))
2106
2107 (defun shr-pixel-buffer-width ()
2108 (if (not shr-use-fonts)
2109 (save-excursion
2110 (goto-char (point-min))
2111 (let ((max 0))
2112 (while (not (eobp))
2113 (end-of-line)
2114 (setq max (max max (current-column)))
2115 (forward-line 1))
2116 max))
2117 (if (get-buffer-window)
2118 (car (window-text-pixel-size nil (point-min) (point-max)))
2119 (save-window-excursion
2120 (set-window-buffer nil (current-buffer))
2121 (car (window-text-pixel-size nil (point-min) (point-max)))))))
2122
2123 (defun shr-render-td (dom width fill)
2124 (let ((cache (intern (format "shr-td-cache-%s-%s" width fill))))
2125 (or (dom-attr dom cache)
2126 (and fill
2127 (let (result)
2128 (dolist (attr (dom-attributes dom))
2129 (let ((name (symbol-name (car attr))))
2130 (when (string-match "shr-td-cache-\\([0-9]+\\)-nil" name)
2131 (let ((cache-width (string-to-number
2132 (match-string 1 name))))
2133 (when (and (>= cache-width width)
2134 (<= (car (cdr attr)) width))
2135 (setq result (cdr attr)))))))
2136 result))
2137 (let ((result (shr-render-td-1 dom width fill)))
2138 (dom-set-attribute dom cache result)
2139 result))))
2140
2141 (defun shr-render-td-1 (dom width fill)
2142 (with-temp-buffer
2143 (let ((bgcolor (dom-attr dom 'bgcolor))
2144 (fgcolor (dom-attr dom 'fgcolor))
2145 (style (dom-attr dom 'style))
2146 (shr-stylesheet shr-stylesheet)
2147 (max-width 0)
2148 natural-width)
2149 (when style
2150 (setq style (and (string-match "color" style)
2151 (shr-parse-style style))))
2152 (when bgcolor
2153 (setq style (nconc (list (cons 'background-color bgcolor))
2154 style)))
2155 (when fgcolor
2156 (setq style (nconc (list (cons 'color fgcolor)) style)))
2157 (when style
2158 (setq shr-stylesheet (append style shr-stylesheet)))
2159 (let ((shr-internal-width width)
2160 (shr-indentation 0))
2161 (shr-descend dom))
2162 (save-window-excursion
2163 (set-window-buffer nil (current-buffer))
2164 (unless fill
2165 (setq natural-width
2166 (or (dom-attr dom 'shr-td-cache-natural)
2167 (let ((natural (max (shr-pixel-buffer-width)
2168 (shr-dom-max-natural-width dom 0))))
2169 (dom-set-attribute dom 'shr-td-cache-natural natural)
2170 natural))))
2171 (if (and natural-width
2172 (<= natural-width width))
2173 (setq max-width natural-width)
2174 (let ((shr-internal-width width))
2175 (shr-fill-lines (point-min) (point-max))
2176 (setq max-width (shr-pixel-buffer-width)))))
2177 (goto-char (point-max))
2178 ;; Delete padding at the bottom of the TDs.
2179 (delete-region
2180 (point)
2181 (progn
2182 (skip-chars-backward " \t\n")
2183 (end-of-line)
2184 (point)))
2185 (goto-char (point-min))
2186 (list max-width
2187 natural-width
2188 (count-lines (point-min) (point-max))
2189 (split-string (buffer-string) "\n")
2190 (if (dom-attr dom 'colspan)
2191 (string-to-number (dom-attr dom 'colspan))
2192 1)
2193 (cdr (assq 'color shr-stylesheet))
2194 (cdr (assq 'background-color shr-stylesheet))))))
2195
2196 (defun shr-dom-max-natural-width (dom max)
2197 (if (eq (dom-tag dom) 'table)
2198 (max max (or
2199 (loop for line in (dom-attr dom 'shr-suggested-widths)
2200 maximize (+
2201 shr-table-separator-length
2202 (loop for elem in line
2203 summing
2204 (+ (cdr elem)
2205 (* 2 shr-table-separator-length)))))
2206 0))
2207 (dolist (child (dom-children dom))
2208 (unless (stringp child)
2209 (setq max (max (shr-dom-max-natural-width child max)))))
2210 max))
2211
2212 (defun shr-buffer-width ()
2213 (goto-char (point-min))
2214 (let ((max 0))
2215 (while (not (eobp))
2216 (end-of-line)
2217 (setq max (max max (current-column)))
2218 (forward-line 1))
2219 max))
2220
2221 (defun shr-pro-rate-columns (columns)
2222 (let ((total-percentage 0)
2223 (widths (make-vector (length columns) 0)))
2224 (dotimes (i (length columns))
2225 (setq total-percentage (+ total-percentage (aref columns i))))
2226 (setq total-percentage (/ 1.0 total-percentage))
2227 (dotimes (i (length columns))
2228 (aset widths i (max (truncate (* (aref columns i)
2229 total-percentage
2230 (- shr-internal-width
2231 (* (1+ (length columns))
2232 shr-table-separator-pixel-width))))
2233 10)))
2234 widths))
2235
2236 ;; Return a summary of the number and shape of the TDs in the table.
2237 (defun shr-column-specs (dom)
2238 (let ((columns (make-vector (shr-max-columns dom) 1)))
2239 (dolist (row (dom-non-text-children dom))
2240 (when (eq (dom-tag row) 'tr)
2241 (let ((i 0))
2242 (dolist (column (dom-non-text-children row))
2243 (when (memq (dom-tag column) '(td th))
2244 (let ((width (dom-attr column 'width)))
2245 (when (and width
2246 (string-match "\\([0-9]+\\)%" width)
2247 (not (zerop (setq width (string-to-number
2248 (match-string 1 width))))))
2249 (aset columns i (/ width 100.0))))
2250 (setq i (1+ i)))))))
2251 columns))
2252
2253 (defun shr-count (dom elem)
2254 (let ((i 0))
2255 (dolist (sub (dom-children dom))
2256 (when (and (not (stringp sub))
2257 (eq (dom-tag sub) elem))
2258 (setq i (1+ i))))
2259 i))
2260
2261 (defun shr-max-columns (dom)
2262 (let ((max 0))
2263 (dolist (row (dom-children dom))
2264 (when (and (not (stringp row))
2265 (eq (dom-tag row) 'tr))
2266 (setq max (max max (+ (shr-count row 'td)
2267 (shr-count row 'th))))))
2268 max))
2269
2270 (provide 'shr)
2271
2272 ;;; shr.el ends here